DNA and RNA Topoisomerase Activities of Top3β Are Promoted by Mediator Protein Tudor Domain-Containing Protein 3

DNA and RNA topoisomerase activities of Top3β are promoted by mediator protein Tudor domain-containing protein 3

Grace Ee-Lu Siawa,b,c,1,2, I-Fen Liuc,1, Po-Yen Linc, Michael D. Beend,2, and Tao-shih Hsiehc,d,3

aMolecular Cell Biology, Taiwan International Graduate Program, Academia Sinica, Taipei 115, Taiwan; bGraduate Institute of Life Sciences, National Defense Medical Center, Taipei 114, Taiwan; cInstitute of Cellular and Organismic Biology, Academia Sinica, Taipei 115, Taiwan; and dDepartment of Biochemistry, Duke University Medical Center, Durham, NC 27710

Edited by James M. Berger, The Johns Hopkins University School of Medicine, Baltimore, MD, and approved July 19, 2016 (received for review April 5, 2016) Topoisomerase 3β (Top3β) can associate with the mediator protein Early experiments involving the knotting of a single-stranded Tudor domain-containing protein 3 (TDRD3) to participate in two circle in trefoil formation (8) and protein-linked RNA cleavage gene expression processes of transcription and translation. De- (9, 10) suggest the biochemical feasibility of this idea. The interest spite the apparent importance of TDRD3 in binding with Top3β is further boosted by recent experiments demonstrating that hu- and directing it to cellular compartments critical for gene expres- man and Drosophila topoisomerase 3β (Top3β) is associated with sion, the biochemical mechanism of how TDRD3 can affect the func- mRNA in cytoplasm and has RNA topoisomerase activity (11, 12). tions of Top3β is not known. We report here sensitive biochemical The deficiency in this enzyme is attributed to one of the causative assays for the activities of Top3β on DNA and RNA substrates in mutations for schizophrenia and cognitive disorder (11, 12) and resolving topological entanglements and for the analysis of TDRD3 results in abnormal synaptic structures in Drosophila (12). functions. TDRD3 stimulates the relaxation activity of Top3β on How RNA or DNA topoisomerase activity can affect neuro- hypernegatively supercoiled DNA and changes the reaction from logical functions remains to be elucidated. However, there are a distributive to a processive mode. Both supercoil retention as- interesting insights to be gained by considering the interacting says and binding measurement by fluorescence anisotropy reveal partners of Top3β. A mediator protein with multiple domains for a heretofore unknown preference for binding single-stranded interprotein interactions, TDRD3 is a key partner for Top3β nucleic acids over duplex. Whereas TDRD3 has a structure-specific in both the nucleus and cytoplasm (13). TDRD3 can recognize binding preference, it does not discriminate between DNA and dimethylated arginine in proteins including inner core histones RNA. This unique property for binding with nucleic acids can have and RNA polymerase II, thus targeting its partner proteins to the an important function in serving as a hub to form nucleoprotein transcriptionally active chromatin, especially at their 5′ end (14, complexes on DNA and RNA. To gain insight into the roles of Top3β on RNA metabolism, we designed an assay by annealing two single- 15). Because the knockdown or deletion of TDRD3 results in stranded RNA circles with complementary sequences. Top3β is ca- R-loop formation near the promoter regions, it is plausible that β pable of converting two such single-stranded RNA circles into a the absence of Top3 near the transcriptional start site can lead to double-stranded RNA circle, and this strand-annealing activity is en- a buildup of excessive negative supercoiling, thus stabilizing the hanced by TDRD3. These results demonstrate that TDRD3 can en- R-loop structure (13). TDRD3 also associates with the partner hance the biochemical activities of Top3β on both DNA and RNA substrates, in addition to its function of targeting Top3β to critical Significance sites in subcellular compartments. Tudor domain-containing protein 3 (TDRD3), a multidomain Tudor domain-containing domain 3 | DNA topoisomerases | scaffold protein functions as an epigenetic reader on nuclear RNA topoisomerases | RNA circles | RNA duplex chromatin and binds with fragile X mental retardation protein on mRNA. It forms a conserved complex with topoisomerase 3β igher-order structural complexities in nucleic acids can be (Top3β), a type IA topoisomerase, and participates in both Hbrought about by the folding and intertwining through inter- transcription and translation. The mechanism of how TDRD3 and intramolecular base pairing. They are impediments to the acts as Top3β’s partner in regulating these cellular processes is transactions of genetic information, including replication, tran- unknown. Here, we demonstrated that TDRD3 is able to stim- scription, recombination, and translation, which involve the un- ulate Top3β’s DNA and RNA topoisomerase catalytic activities, winding and rewinding of base-paired regions to access encoded through binding and stabilizing single-stranded regions in DNA information (1–3). DNA topoisomerases are nature’s tools to and RNA substrates. Because these regions are the preferred resolve the topological entanglements in nucleic acids. They are site for Top3β, TDRD3 therefore acts as a regulator to provide ubiquitous in nature, first discovered in bacteria in 1971 (4), and access of the enzyme to these nucleic acid substrates and to act in mammalian cells in 1972 (5), and are characterized by a upon them. mechanism involving the formation of a covalent protein/DNA adduct to generate a transient and reversible break allowing for Author contributions: G.E.-L.S., I.-F.L., and T.-s.H. designed research; G.E.-L.S., I.-F.L., and P.-Y.L. performed research; G.E.-L.S., I.-F.L., P.-Y.L., M.D.B., and T.-s.H. contributed new topological transformation (6). Based on whether the strand pas- reagents/analytic tools; G.E.-L.S., I.-F.L., M.D.B., and T.-s.H. analyzed data; and G.E.-L.S., I.-F.L., sage is through a protein-mediated single-stranded gate or a dou- M.D.B., and T.-s.H. wrote the paper. ble-stranded gate, the enzymes are classified into type I or type II The authors declare no conflict of interest. topoisomerases, respectively. Both types are further classified into This article is a PNAS Direct Submission. A and B families based on their structural and mechanistic features 1G.E.-L.S. and I.-F.L. contributed equally to this work. (7). All topoisomerases are essential for the growth and develop- 2To whom correspondence may be addressed. Email: [email protected] or graceelsiaw@ ment of an organism, suggesting that they have critical but distinct gmail.com. functions in all cells. 3Deceased August 4, 2016. Ever since their discovery, there is a fascination for whether This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. these critical DNA enzymes may play a role in the RNA world. 1073/pnas.1605517113/-/DCSupplemental.

E5544–E5551 | PNAS | Published online August 31, 2016 www.pnas.org/cgi/doi/10.1073/pnas.1605517113 Downloaded by guest on October 2, 2021 protein fragile X mental retardation protein (FMRP) in the PNAS PLUS cytoplasm. Missing FMRP can cause fragile X mental retardation, a major cause of congenital mental disorder (16). FMRP is preferentially localized in the coding regions of mRNA and plays a role in inhibiting the translation of mRNA relevant to synaptic functions in neuronal cells (17). It is thus suggested that Top3β may also have a role in translation through its RNA topoisomerase activity in modulating the complex structures of folded mRNA (12). Whereas Top3β can have important functions in two processes of gene expression in nucleus and cytoplasm through its interactions with TDRD3, it is unclear whether TDRD3 has any additional role in regulating Top3β activity, beyond its targeting function of directing Top3β to critical sites in subcellular compartments. Our results presented here show that TDRD3 can promote both DNA and RNA topoisomerase activities, likely through stabilizing the complex of the enzyme bound to the single- stranded regions of the substrate, which is also the site where type IA enzymes carry out strand passage reactions. The finding that TDRD3 has a structure-specific preference for single- stranded nucleic acids, but does not discriminate between DNA and RNA, adds an aspect to the scaffolding function of TDRD3 in the assembly of the nucleoprotein complex. A mediator protein that is able to interact with specific proteins and to target to the preferred structure of nucleic acids can have a versatile role serving as a hub for assembling stable complexes in chromatin or polysomes.

With an increasing interest in RNA topoisomerase activity, it BIOCHEMISTRY is important to develop biochemical assays other than the trefoil formation of circular RNA. Using two single-stranded circular RNA substrates with complementary sequences, we have demonstrated that Top3β and other type IA enzymes can mediate the formation of double-stranded circular RNA through the in- termolecular stranded passage activity of these enzymes. This circle-annealing assay poses a less stringent requirement for the sequence and structure of the substrate RNA compared with the Fig. 1. TDRD3 stimulates the relaxation activity of Top3β on hypernegatively knotting assay, and can thus broaden our capability to test the effect supercoiled DNA substrates. Topological states of the plasmid DNA were: of an RNA sequence or structure on RNA topoisomerase ac- hypernegatively supercoiled (H), negatively supercoiled (N), relaxed circle (R), tivity. In addition, this versatile biochemical method to prepare nicked circle (Ni), and linearized (L). (A) Titration assay of TDRD3 concentra- β an interesting class of molecules of circular duplex RNA has tions showed that relaxation activity of Top3 was stimulated in both a time- unique potential for investigating the functions of other RNA and concentration-dependent manner. (B) Time course assay showed that relaxation by Top3β alone occurs in a distributive manner, and the presence enzymes, including topoisomerases. of TDRD3 in the reactions changes the relaxation mode to a processive one. (C) Top3β- and TDRD3-catalyzed relaxation activity was temperature Results dependent. Stimulation of Top3β Relaxation Activity by TDRD3. A sensitive method for detecting the effect of TDRD3 on Top3β’s activity is the relaxation of hypernegatively supercoiled DNA by Drosophila relaxation, wherein intermediates precede the appearance of the Top3β, where the DNA is relaxed to a supercoiling state that is final product. In contrast, in reactions with both Top3β and slightly more negatively supercoiled than the plasmid DNA (18). TDRD3, the relaxation of Top3β changed to a processive mode In reactions with 7.5 nM Top3β, increasing the concentration of without the accumulation of intermediates. It is plausible that TDRD3 stimulated the conversion of hypernegatively super- TDRD3 can enhance the binding of Top3β to the DNA substrate coiled substrate to a final product with a mobility just behind the and allow the reaction to proceed to completion before releasing negatively supercoiled plasmid DNA (Fig. 1A). When the Top3β its binding to DNA. concentration was doubled, an approximately twofold higher Because Top3β, like other type IA enzymes, prefers DNA concentration of TDRD3 was required to duplicate the same substrates with underwound regions, the effect of TDRD3 in stimulation pattern, suggesting stoichiometry-dependent stimu- promoting the binding of Top3β to DNA may be mediated latory effect of TDRD3. In both cases, reactions became satu- through this structure. We tested this hypothesis by examining rated, but without the DNA reaching a fully relaxed state, when whether its stimulatory effect is also favored under conditions TDRD3 and Top3β reached a similar stoichiometric ratio. The promoting the generation of underwound regions in DNA. In- data suggest a possible protein–protein interaction between deed the stimulatory effect of TDRD3 is more apparent when Top3β and TDRD3. the reactions were carried at 37 °C, 42 °C, and 50 °C, but not at To gain insight into the mechanistic basis for stimulatory effect lower temperatures (Fig. 1C), indicating that an increase in the of TDRD3, we examined the kinetic behavior of Top3β re- reaction temperature promotes the stimulatory effect of TDRD3. laxation with and without TDRD3 (Fig. 1B). With Top3β alone, Complete substrate relaxation was seen when incubation was intermediate products consisting of DNA molecules with dif- performed at 37 °C and 42 °C, with an incomplete reaction at ferent supercoiling densities were observed at early time points 50 °C possibly due to partial enzyme inactivation at this elevated before the reaction was completed with the appearance of the temperature. Therefore, one possible mechanism for how TDRD3 final product. Thus, the enzyme showed a distributive mode of regulates the activity of Top3β is through binding and stabilizing

Siaw et al. PNAS | Published online August 31, 2016 | E5545 Downloaded by guest on October 2, 2021 single-stranded regions in plasmid substrates, and thereby recruits numbers due to the presence of ethidium and the effect of the the enzyme to these sites. addition of TDRD3. Analysis of identical samples with gel electrophoresis without chloroquine supports the notion that TDRD3 Retention of Negative Supercoiling by TDRD3 in Plasmid DNA. Given retains DNA supercoiling and that the addition of ethidium fur- the proposed mechanism that TDRD3 stimulates Top3β’s ac- ther introduces more negative supercoiling to these DNAs (Fig. tivity through its binding to the single-stranded region in the S2). These results suggest that Top1 can efficiently remove DNA plasmid DNA, one would expect it to retain negative supercoiling supercoiling regardless of the presence of TDRD3. in the presence of a topoisomerase to remove unconstrained To further demonstrate the potential supercoiling retention supercoiling. We examined whether TDRD3 could alter local ability of TDRD3, we first used Top1 to relax the DNA, and then DNA topology by using two approaches to assay the negative compared the supercoiling differences after incubating with or supercoiling retention (for schematic diagrams of the design of without TDRD3. The idea is that because there was no super- these experiments, see Fig. S1 A and B). In these experiments, we coiling in DNA before the addition of TDRD3, any change of used Drosophila topoisomerase I (Top1), a type IB enzyme, which supercoiling must be due to the effect of TDRD3 (Fig. S1B). provides an efficient swivel to remove unconstrained supercoiling This experiment can minimize the possible inhibitory effect of in the circular DNA substrates. We first used negatively super- TDRD3 on Top1 relaxation activity. However, because the coiled plasmid pUC19 DNA as a substrate to demonstrate the starting DNA substrate is relaxed and contains very few under- retention of plasmid negative supercoiling by TDRD3, as shown wound regions, one would expect that the supercoiling retention by the reduction in linking numbers of the final products, in by TDRD3 does not reach the same extent as was observed with contrast to control reactions with Top1 only (Fig. 2A). However, negatively supercoiled DNA (Fig. 2A). This is born out in the an alternative interpretation would be that the retention of experiments showing the dose-dependent retention of negative negative supercoiling in the presence of TDRD3 was due to its supercoiling (Fig. 2B). The DNA unwinding is at much lower inhibition of Top1 relaxation activity. To demonstrate that Top1 levels compared with results using plasmid DNA as a starting remained active and capable of removing supercoiling in reactions substrate. These results demonstrate that, whereas TDRD3 containing TDRD3, we added the DNA intercalator ethidium cannot actively unwind DNA, it can bind with higher affinity at bromide (EtBr) to these reactions to introduce a dose-dependent the underwound region in the plasmid DNA. Through its known β β unwinding of DNA. Additional negative supercoiling due to the interaction with Top3 , TDRD3 can thereby target Top3 to its presence of ethidium was observed in the reactions with TDRD3. preferred DNA structure and carry out strand passage reaction. These reaction products were analyzed by gel electrophoresis with chloroquine to help reveal the relative linking deficiencies among Preference of TDRD3 in Binding Single-Stranded Nucleic Acids. The the samples, thus demonstrating the gradual reduction of linking retention of negative supercoiling assayed here by the topological shift method supports the idea that TDRD3 can preferentially associate with a single-stranded region of DNA, it does not rule out the possibility of other topological changes such as left- handed DNA wrapping associated with protein binding. We therefore directly monitored the binding affinities of TDRD3 with single- vs. double-stranded nucleic acids using fluorescence anisotropy to detect the binding to DNA and RNA oligomers tagged with a fluorophore. The temperature of the reactions was reduced from 37 °C to 25 °C to minimize fraying of double- stranded oligonucleotides. We observed anisotropy changes that were associated with the binding of TDRD3 to distinct structures of fluorescein-labeled DNA and RNA oligomers (Fig. 3 A and B). A single-stranded 73-nucleotide DNA with mixed sequence showed both a larger δ-anisotropy and a lower K1/2 (the protein concentration giving half-maximal binding) than did the same sequence in duplex form, indicating a higher affinity of TDRD3 for the single-stranded form (Fig. 3A). By the same criteria, the homopolymeric sequence (dT)73, showed higher affinity with TDRD3 than did the single-stranded DNA with a mixed sequence. When tested with an RNA oligonucleotide of similar sequence and length, we observed the same order of preference for TDRD3 binding with (rU)73 having the highest affinity and duplex RNA having the lowest (Fig. 3B). Fluorescence anisotropy experiments therefore indicated a preference for TDRD3 binding to nucleic acids with the least secondary structure, whether they are DNA or RNA. These results demonstrating that TDRD3 appears to bind RNA as well as it does to DNA, together with recent work that showed human Top3β possesses RNA topoisomerase activity (12), Fig. 2. TDRD3 demonstrates retention of negative supercoiling in plasmid led us to investigate in more detail the effect of TDRD3 on the substrates. (A) Addition of TDRD3 generates final products with greater RNA topoisomerase activity of Top3β. linking number deficiency than control reactions without TDRD3. The extent of negative supercoiling retention was manifested by using different con- Synthesis of Single-Stranded RNA Circles Through Self-Splicing of centrations of EtBr. Final reaction products were analyzed in the presence a Group I Intron Sequence and Its Use as a Substrate for RNA of chloroquine to distinguish differences in negative supercoiling densities Topoisomerase. High binding affinity to single-stranded RNAs by (−, reaction was stopped immediately after Top1 incubation). (B) TDRD3 TDRD3 suggests that it may also play a role in Top3β-mediated generates final products with linking number deficiency from Top1-relaxed Escherichia pUC19 DNA, compared with control reactions with Top1 only. The extent of activity on the RNA substrate. Two groups have shown supercoiling retention was dependent on TDRD3 concentrations. (N, nega- coli Top3 and human Top3β have RNA topoisomerase activity tively supercoiled substrates; T3, 2 μM TDRD3 without Top1.) capable of interconverting single-stranded RNA circles to knots

E5546 | www.pnas.org/cgi/doi/10.1073/pnas.1605517113 Siaw et al. Downloaded by guest on October 2, 2021 PNAS PLUS

Fig. 3. TDRD3 shows preference to bind single-stranded DNA and RNA oligomers. (A) TDRD3 binds to 73-bp duplex DNA, 73-nucleotide ssDNA, and (dT)73 oligomers with apparent increasing affinity; half-maximal binding (K1/2) under these conditions was seen with ∼300 nM, 110 nM, and 70 nM TDRD3 for duplex DNA, ssDNA, and the dT oligomers, respectively. (B) The binding of TDRD3 to RNA oligomers showed a pattern similar as the DNA substrates, with higher

affinity for (rU)73 oligomers compared with ssRNA oligomers. Very little binding to duplex RNA was detected. Half-maximal binding was seen with ∼190 nM for ssRNA, and 90 nM TDRD3 for the rU oligomer. The average and the SDs of δ-anisotropy values for each TDRD3 concentration were calculated from triplicate measurements (n = 3).

(8, 12), but this activity depends on a substrate sequence enabling single-stranded RNA circles with complementary sequences. A BIOCHEMISTRY a folded secondary structure. To gain insight into the RNA top- 676-nt-long precursor RNA was observed after a 1 h transcrip- oisomerase activity catalyzed by Top3β, we prepared completion and various spliced products were detected after a 20-h in- mentary single-stranded RNA circles that could be annealed to cubation (Fig. S3B). Candidate bands from each construct were form a substrate with a linking number of zero. Topoisomerase isolated from 3.5% denaturing polyacrylamide gel (Fig. S3B)and activity on this substrate would be predicted to generate linked further characterized by mild alkaline degradation, RNase R treat- RNA circles. The single-stranded circles were generated with ment, and gel mobility with respect to markers. Alkaline treatment a reengineered self-splicing group I intron (19, 20) containing the of gel-purified circular RNA candidates (+cRNA and −cRNA) sequence of a permuted intron–exon (PIE) version of the Anabaena produced fragments with a discrete shift in mobility (Fig. 4A), in + pre-tRNA group I intron (21). Two constructs, PIE–insert380 contrast to a smear of fragments produced from the linear form − and PIE–insert380 , contain a 380-bp fragment in opposite ori- (LRNA). The nicked product of each circular RNA migrates entation flanked by the autocatalytic splice sites (Fig. S3A). The close to the position of the corresponding LRNA marker. We in vitro transcripts from these two constructs can generate two further used a 3′-to-5′ exoribonuclease, RNase R, specific for the

Fig. 4. Splicing of PIE–insert380 and identification of single-stranded RNA circles. (A) Single-stranded circular (+cRNA or −cRNA) and LRNA were isolated from spliced products (P1 and P2). Gel-purified circular RNA candidates were treated in nicking conditions (N) or RNase R digestion (R) to validate the circular structure. With partial alkali hydrolysis, cRNA produced predominantly LRNA, compared with the smear pattern produced from LRNA. The resistance to RNase R digestion confirmed the nonlinear structure of +cRNA and −cRNA. Samples without treatments were shown as controls (−). (B) cRNA showed a larger mobility difference relative to markers (M) between 3.5% (Left) and 5% (Right) denaturing polyacrylamide gels, verifying the predicted circular structure. Control LRNA showed similar mobility on both gels.

Siaw et al. PNAS | Published online August 31, 2016 | E5547 Downloaded by guest on October 2, 2021 degradation of linear RNAs, to confirm its circularity. Both +cRNA and −cRNA are resistant to RNase R, whereas LRNA is completely degraded (Fig. 4A). Lastly, the electrophoretic mobility of cRNAs is expected to be more sensitive to the gel concentration than LRNA. With electrophoresis on 3.5% and 5% denaturing gels, cRNA has a much slower migration on a higher percentage (5%) gel, but the control LRNA shows mobility similar to the 400-nt marker of RNA ladder on both gel conditions (Fig. 4B). These in-vitro–generated single-stranded RNA circles validated by three independent methods are now useful as a substrate for monitoring RNA topoisomerase activity.

Effect of TDRD3 on RNA Strand-Annealing Activity by Top3β. To date, the only method to assay the catalytic activity of RNA topoisomerase is to detect the interconversion of an open circle and trefoil knot (8). In this study, we designed another assay by monitoring strand annealing between two complementary single- stranded RNA circles (ss cRNA). By analyzing the reaction products from incubating RNA circles with Top3β and/or TDRD3 in a denaturing gel, we showed that Top3β is capable of generating a product band with slower migration compared with the substrates (Fig. 5A). The structure of this product was later identified as a double-stranded RNA circle (ds cRNA) by electron microscopy (Fig. 6). Moreover, the presence of TDRD3 can enhance Top3β’s activity, resulting in an increase in the formation of a double- stranded RNA circle. The active site mutant Top3β-Y332F failed to generate any products, indicating that annealing of single-stranded RNA circles depends on the strand passage activity of Top3β.The stimulation of RNA topoisomerase activity of Top3β by TDRD3 is dose dependent, with the stoichiometry reaching a plateau near 2:1, similar to its effect on DNA topoisomerase activity (Fig. 5B). To test if this RNA annealing activity can be used to examine the RNA topoisomerase activity in other type IA enzymes, we tested with the prototype type IA topoisomerase, E. coli Top1 (EcTop1) and the newly identified hyperthermophile Nano- archaeum equitans Top3 (NeqTop3). Both NeqTop3 and EcTop1 have ability to generate double-stranded RNA circles (Fig. S4 A and B), and again the active site mutants (EcTop1-Y319F and β NeqTop3-Y293F) showed no activity. TDRD3 has no stimulatory Fig. 5. TDRD3 stimulates RNA strand-annealing activity catalyzed by Top3 . ’ (A) Single-stranded +cRNA and −cRNA were incubated with 0 (−)or90nM effect on EcTop1 s activity, indicating that the specificity of in- β β β Top3 . Top3 catalyzes the conversion of single-stranded cRNA into double- teraction between TDRD3 and Top3 may be critical for its stim- stranded cRNA, and the presence of 90 nM TDRD3 enhanced Top3β’s RNA ulation (Fig. S4C). RNA products generated by Top3β and topoisomerase activity. The active site mutant (Top3β-Y332F) showed no NeqTop3 were resistant to RNase R digestion under conditions in strand annealing activity. LRNA is a breakdown product of single-stranded which linear RNA was completely degraded (Fig. S5), confirming RNA circles. (B) Titration of TDRD3 concentrations showed a dose-dependent that the RNA strands remained intact and circular after treatment. stimulation in RNA strand-annealing activity catalyzed by Top3β.

Direct Visualization of Double-Stranded RNA Circles by Electron Microscopy. The structures of the circular RNA substrate and are 124 nm and 128 nm, respectively, comparable to the estimated topoisomerase-mediated annealing products were further char- length of 113 nm based on a molecule with 403 bp and the known acterized by EM (Fig. 6A). The 403-nt-long single-stranded RNA helical rise of A-RNA being 0.28 nm (22, 23). In addition to circles appeared as discrete collapsed blobs, presumably reflecting double-stranded RNA circles, Top3β was able to generate a higher the formation of secondary structures in RNA. In the presence of molecular weight RNA product upon TDRD3 stimulation (Fig. single-stranded nucleic acid binding protein (SSB), the single- 5). The structure of this product determined by EM appeared as stranded RNA-containing nucleoprotein complex appeared as oligomers of double-stranded RNA circles (Fig. S7), indicating small circles. The presumptive double-stranded circular RNA that Top3β has the ability to conjoin two or more double-stranded product generated by Top3β was purified from denaturing gels RNA circles to form interlinked multirings. Our results showed and visualized by cytochrome c spreading EM. Larger circles were that TDRD3 can efficiently bind to single-stranded DNA and detected with a contour length expected for double-stranded RNA RNA, and may be recruited to the junction of single-stranded (Fig. 6A and Fig. S6A). These data are consistent with the con- and double-stranded regions. For the double-stranded RNA clusion that Top3β is able to convert single-stranded RNA circles circles produced here, there is an expected 7-nt single-stranded into a double-stranded form. We also examined the RNA bubble because the same splice site junction sequence is pre- products generated by NeqTop3, and observed that RNA circles sent in both senses of single-stranded circular RNA (Fig. S3). generated by NeqTop3 are similar to those from Top3β (Fig. We therefore incubated double-stranded RNA circles with S6B). We measured the contour length of RNA samples, and TDRD3, and visualized the RNA–protein complex by direct- that from pUC19 DNA as the length marker for comparison, mounting EM. We found that TDRD3 binds to a single site in an with the circumference distribution shown in Fig. S6 A and B. RNA circle, and also sometimes between two of them (Fig. 6B). The measured contour length of pUC19 with 2,686 bp is 917 nm, The direct examination by EM further demonstrated that Top3β and those for dsRNA circles generated by Top3β and NeqTop3 possesses an RNA topoisomerase activity to anneal two circular

E5548 | www.pnas.org/cgi/doi/10.1073/pnas.1605517113 Siaw et al. Downloaded by guest on October 2, 2021 PNAS PLUS BIOCHEMISTRY

Fig. 6. Visualization of RNA circles by electron microscopy. (A) EM images of single-stranded RNA circles (collapsed structures), partially opened single-stranded RNA circles bound by SSB, and double-stranded RNA circles generated by Top3β,shownLeft, Middle,andRight, respectively. The molar ratio of ss cRNA to SSB is 1:65. (B) Double-stranded RNA circles bound by TDRD3. The molar ratio of ds cRNA to TDRD3 is 1:10. The Left image shows ds cRNA without protein. (Scale bars, 50 nm.)

molecules, and this activity is promoted by TDRD3 via its asso- can enhance Top3β relaxation activity and shift the reaction from ciation with single-stranded regions in the substrate, thus pro- a distributive to processive mode, suggesting that it can stabilize viding access to Top3β. the enzyme-bound DNA complex until the excessive supercoiling is removed. This effect is likely caused by the preference of Discussion TDRD3 to bind single-stranded nucleic acids, also the known TDRD3 is a mediator protein known to participate in multiple binding site for a type IA topoisomerase like Top3β. The nucleoprotein complexes through its interactions with key cel- N-terminal domain of TDRD3 harbors an intact oligonucleotide/ lular proteins. It can function in two cellular compartments, the oligosaccharide binding (OB) fold (24, 25). This motif has been nucleus and the cytoplasm, regulating gene expression via tran- proposed to bind single-stranded DNA/RNA or function as a scription and translation (14). TDRD3 can bind to the dime- protein-interacting interface (27, 28). Interestingly, this region thylated arginine present in core histones and RNA polymerase had been shown to interact with Top3β (11–13); therefore, it is II, allowing it to target to the actively transcribed chromatin, plausible that the OB fold of TDRD3 is responsible for stimulating especially near the promoter region (14, 15). TDRD3 can also the enzyme’s activities. Whereas TDRD3 is able to bind many key interact with FMRP and target to mRNA (24, 25). Top3β is one of proteins through its OB, ubiquitin-associated (UBA), and Tudor the TDRD3-interacting proteins (11–13). However, its effect on domains (12, 13, 24, 25), the structure specificity in nucleic acids the biochemical activities of topoisomerase has not been thor- binding demonstrated here enables TDRD3 to serve more effec- oughly investigated. We show here that TDRD3 can promote both tively as a scaffold for assembling nucleoprotein complexes. the DNA and the RNA topoisomerase activities of Top3β. TDRD3 can associate with the promoter region of the actively We used relaxation of hypernegatively supercoiled DNA as a transcribed genes and target Top3β to resolve R-loop structures sensitive and facile assay for Top3β. Earlier work demonstrated associated with the movement of a transcriptional fork (13). Be- that Top3β has modest relaxation activity toward plasmid DNA cause the R loop can expose a single-stranded DNA region, it is a (26), while maintaining robust activity for the hypernegatively structure that may facilitate the recruitment of the TDRD3/Top3β −/− supercoiled DNA (18), presumably because of the presence of complex. Depletion of TDRD3 in Tdrd3 primary mouse em- underwound single-stranded regions that serve as the preferred bryonic fibroblast cells can result in DNA strand breaks and the binding site for Top3β. With this assay, we showed that TDRD3 formation of γ-H2AX foci (13). Because single-stranded DNA

Siaw et al. PNAS | Published online August 31, 2016 | E5549 Downloaded by guest on October 2, 2021 regions are known to be vulnerable to DNA damage and strand Our results support the notion that TDRD3 and Top3β can breaks, the recruitment of TDRD3 to this region may help as- function on chromatin in nuclei and on mRNA in cytoplasm. semble a repair machinery for maintaining genome stability. TDRD3 may promote these two activities by targeting the single- More than half of the cellular TDRD3 is localized in the cy- stranded region in nucleic acid substrates, which is the preferred toplasm (24), where it can associate with FMRP, and is present site for Top3β to carry out strand passage. in the stress granules that are the storage compartment for With the unique RNA topoisomerase reaction described here, mRNA during cellular stress (24, 25). TDRD3 can also associate we have the possibility of generating double-stranded circular with Top3β in cytoplasm, and the ternary complex of Top3β/ RNA molecules incorporating any desired sequence in these TDRD3/FMRP can be localized to stress granules and mRNA in molecules. The presence of double-stranded RNA circles has not polysomes (11, 12). Because circularized messenger ribonucleo- been detected in nature. However, recent results indicate that proteins (mRNPs) were found in stress-induced cellular com- single-stranded RNA circles can be made from sequences of partments (29), Top3β and TDRD3 might be involved in multiple genomic loci (30–32). Whereas the exact molecular compacting this complex through topological alterations of function of the circular RNAs remains to be elucidated, there is mRNA. Nevertheless, the biochemical function of Top3β/ a growing awareness of the ubiquitous nature of the presence of TDRD3/FMRP ternary complex remains a question of intense circular RNA and its functional importance. There is a pre- interest. Top3β deficiencies have been linked to schizophrenia ponderance of circular RNA in neuronal cells, and furthermore, and cognitive impairment (11) and to abnormal synaptic struc- some of these RNAs have specific intracellular location (33–35). tures in Drosophila (12). Because Top3β has been shown to The proposed molecular functions of circular RNAs include possess RNA topoisomerase activity, this ternary complex may sponging microRNA and interacting with mRNA for the regu- regulate the translation of long mRNA in neuronal cells through lation of gene expression (36–39). The association and dissocia- modulating the structural complexity of RNA. Interestingly, our tion of these RNAs with a circular counterpart can be facilitated data here also showed that TDRD3 can function beyond being a by an RNA topoisomerase, especially if the partner RNA is long mediator for binding to partner proteins. TDRD3 is a nucleic and has an extensively folded structure. TDRD3-stimulated, acid binding protein that does not discriminate between DNA strand-annealing activity of Top3β might be involved in this func- and RNA and has a strong preference for single-stranded tion, although the mechanism on how the proteins will be recruited structure. This unique binding preference may enable TDRD3 to and how the binding to complex RNA structures occur remain play critical roles in controlling the structure of chromatin and unknown. Facile methods to generate circular single-stranded and β mRNA through its interaction with Top3 . Its single-stranded double-stranded RNA may provide avenues to investigate the specificity can allow TDRD3 to target to the cellular compart- molecular functions of these regulatory molecules. ments that require its presence: the promoter region of actively transcribed chromatin and long mRNA being translated by the Materials and Methods β passage of ribosomes. In addition, Top3 as a type IA enzyme Plasmid Relaxation Assays of Drosophila Top3β and TDRD3. Hypernegatively also prefers to carry out its strand passage action at the single- supercoiled pUC19 DNA (300 ng) were incubated with the indicated amount

stranded region, and in doing so, its activity can be enhanced of Top3β and TDRD3 in 40 mM Hepes-KOH pH 7.5, 1 mM MgCl2,and50μg/mL through interaction with TDRD3. BSA at 37 °C (18) for 60 min. Reactions were stopped by adding to final There has been continuing interest in addressing the question concentrations of 15 mM EDTA, 0.3% SDS, and 1.0 mg/mL proteinase K, of whether a DNA topoisomerase can also function as an RNA heated at 45 °C for 30 min. Electrophoresis with 1.0% Tris/borate/EDTA (TBE) enzyme, and if so, how these two activities are regulated. RNA agarose gel was carried out in the presence of 0.75 μg/mL EtBr for 18 h at 30 V. topoisomerase activity has been demonstrated in E. coli Top3 Drosophila and human Top3β through the knotting of an RNA circle (8, 12) Negative Supercoiling Retention Assays of TDRD3. Two approaches and in vaccinia Top1, human Top1, and E. coli Top3 with cleavage were used here with either supercoiled DNA (i) or relaxed DNA (ii)asthe binding substrate. Ionic buffer conditions were based on previous studies assays (9, 10). Because the trefoil formation assay requires the (18, 40, 41), with modifications. (i) A total of 300 ng negatively supercoiled specific structure of two pairs of intramolecular base-paired re- pUC19 DNA was incubated with 1 μM TDRD3 in 40 mM Hepes-KOH

gions in the RNA circle, it will be important to develop an al- pH 7.5, 5 mM MgCl2,and50μg/mL BSA at 45 °C for 5 min, followed by ternative approach with a less stringent sequence requirement. incubation at 37 °C for 10 min after adding 10 nM Top1. Reactions were either We demonstrated here the application of the group I intron self- stopped or allowed to proceed by adding either 5 or 25 ng EtBr. Following a splicing sequence to generate an RNA circle and the intermo- 10-min incubation at 37 °C, reactions were stopped and phenol was extracted lecular annealing of two circles with complementary sequences to separate EtBr and protein from DNA. (ii) A total of 5 nM Top1 with 300 ng to form double-stranded circles. Because the rewinding of two negatively supercoiled pUC19 DNA was incubated in the presence of 10 mM · μ circular molecules depends absolutely on the presence of an Tris HCl pH 7.9, 50 mM KCl, 10 mM MgCl2, 0.1 mM EDTA, and 50 g/mL BSA for efficient swivel with strand passage activity, it provides a robust 5 min at 37 °C, followed by reactions at 45 °C for 10 min with TDRD3 at the indicated concentrations up to 2 μM. Reactions were stopped as mentioned and sensitive assay for RNA topoisomerase. With this assay, we above and electrophoresed on 1.0% TBE agarose gel with or without 30 μM detected RNA enzyme activity in DNA topoisomerases isolated chloroquine at 30 V for 18 h. from diverse sources, including E. coli Top1, N. equitans Top3, Drosophila β Drosophila β and Top3 . Interestingly, with Top3 ,we Fluorescence Anisotropy Assays of Drosophila TDRD3 Binding to Nucleic Acids. could also observe stimulation of RNA circle annealing activity The 5′-carboxyfluorescein (FAM)-labeled 73 single-stranded DNA and similar RNA by TDRD3. The stimulation of DNA and RNA topoisomerase oligomers (ssDNA and ssRNA), the respective unlabeled complementary strands, activities by TDRD3 is specific to Top3β, but not other type IA dT and rA oligomers were purchased from Integrated DNA Technologies. The enzymes such as N. equitans Top3 and E. coli Top1. A possible sequence of FAM-labeled 73 ssDNA was 5′-GGACTCTGCCTCAAGACGGTAGTCA- mechanism proposed here is, following the binding of TDRD3 to ACGTGACCAGCACTGACCCATTAGGGACCGTCCACCCGACGCCTCCTG-3′. dsDNA Top3β and its nucleic acids substrates via the conserved OB-fold and dsRNA oligomers were generated by annealing respective complementary domain, TDRD3 might be able to stabilize Top3β-bound nucleic ssDNA or ssRNA strands. Anisotropy measurements were performed using acid complex that subsequently stimulates the enzyme’s activity. spectrofluorometer Fluorolog-3 (Horiba Jovin Yvon) at 25 °C for 5 min, with 50 nM substrates, titrated against various concentrations of TDRD3 up to 1 μM, Alternatively, TDRD3 might be competing with other type IA in 10 mM Tris·HClpH7.9,50mMKCl,10mMMgCl, and 0.1 mM EDTA. The enzymes to bind to single-stranded nucleic acids, leading to the 2 average and SD of δ-anisotropy values (r − ro) from triplicate measurements for inhibition of the enzymes’ activity. The prototype type IA top- each TDRD3 concentration were plotted to show the binding pattern of each oisomerase, E. coli Top1 demonstrates this phenomenon, par- nucleic acids substrates to the protein. TDRD3 concentrations that gave half-

ticularly in reactions with relatively high TDRD3 concentrations. maximal binding (K1/2) were estimated both graphically and by fitting to the Hill

E5550 | www.pnas.org/cgi/doi/10.1073/pnas.1605517113 Siaw et al. Downloaded by guest on October 2, 2021 equation for cooperative binding. Similar values were obtained from both shadowing. For visualization of TDRD3 and ds cRNA, purified ds cRNA were PNAS PLUS methods. The values in the legend to Fig. 3 are obtained from curve fitting. incubated with TDRD3 at a molar ratio of 1:10 in 40 mM Hepes-KOH (pH 7.5)

and 1 mM MgCl2 at 37 °C for 20 min, fixed by 0.4% glutaraldehyde at room RNA Strand-Annealing Assays. The assay contained 15 ng of single-stranded temperature for 5 min, followed by adding a final concentration of 100 mM

+cRNA and –cRNA, 40 mM Hepes-KOH (pH 7.5), 5 mM MgCl2,50μg/mL BSA, Tris·HCl (pH 7.5) to terminate the fixation. The RNA–protein complex was in- 7.5 mM DTT, and Top3β or TDRD3 with indicated concentrations. Reactions cubated with spermidine buffer (2 mM spermidine HCl, 10 mM Tris·HCl (pH 7.5),

were performed at 37 °C for 1 h and stopped with 10 mM EDTA, 0.2% SDS, 50 mM NaCl, 75 mM KCl, 2 mM MgCl2, and 0.5 mM CaCl2), mounted directly and 1 mg/mL proteinase K at 45 °C for 30 min. For EcTop1 and EcTop1-Y319F on glow-discharged carbon-coated 400-mesh copper grids (EM Sciences), then · mutant, reactions were performed in 10 mM Tris HCl (pH 8.0), 50 mM NaCl, washed with a water/ethanol series, followed by rotary shadowcast with μ 100 g/mL gelatin, and 6 mM MgCl2 at 37 °C for 1 h. Reactions for NeqTop3 tungsten (43). Grids were imaged with a FEI Tecnai G2 F20 TEM instrument at and NeqTop3-Y293F mutant were incubated at 56 °C for 0.5 h in 10 mM 120 kV and a Gatan US1000 digital camera. To visualize ss cRNA, RNA was Tris·HCl (pH 8.0), 50 mM KCl, 10 mM MgCl , 0.1 mM EDTA, and 50 μg/mL 2 incubated with extreme thermostable SSB (New England Biolabs) at 50 °C for gelatin. All samples were denatured with formamide in loading buffer (95% 20 min followed by fixation, gel filtration, and direct mounting. formamide, 18 mM EDTA, 0.025% SDS, and xylene cyanol) at 90 °C for 4 min and run on 3% TBE-urea polyacrylamide gels at 150 V for 1 h followed by ACKNOWLEDGMENTS. We thank Dr. Yuk-Ching Tse-Dinh (Florida Interna- SYBR Gold staining (Invitrogen). tional University) for wild-type and mutant (Y319F) EcTop1 proteins and Hung-Hsun Shuai for technical assistance. This work was supported by Electron Microscopy. Gel-purified double-stranded RNA circles were examined Ministry of Science and Technology Taiwan Grant MOST-104-2321-B-001-012 by the cytochrome c drop-spreading method (42) and processed for tungsten and Academia Sinica intramural funding AS-105-TP-B04 (to T.-s.H.).

1. Wang JC (2009) Untangling the Double Helix. DNA Entanglement and the Action of 25. Linder B, et al. (2008) Tdrd3 is a novel stress granule-associated protein interacting the DNA Topoisomerases (Cold Spring Harbor Lab Press, Cold Spring Harbor, NY). with the Fragile-X syndrome protein FMRP. Hum Mol Genet 17(20):3236–3246. 2. Wang JC (2002) Cellular roles of DNA topoisomerases: A molecular perspective. Nat 26. Seki T, Seki M, Onodera R, Katada T, Enomoto T (1998) Cloning of cDNA encoding a Rev Mol Cell Biol 3(6):430–440. novel mouse DNA topoisomerase III (Topo IIIbeta) possessing negatively supercoiled 3. Chen SH, Chan NL, Hsieh TS (2013) New mechanistic and functional insights into DNA DNA relaxing activity, whose message is highly expressed in the testis. J Biol Chem topoisomerases. Annu Rev Biochem 82:139–170. 273(44):28553–28556. 4. Wang JC (1971) Interaction between DNA and an Escherichia coli protein omega. 27. Theobald DL, Mitton-Fry RM, Wuttke DS (2003) Nucleic acid recognition by OB-fold J Mol Biol 55(3):523–533. proteins. Annu Rev Biophys Biomol Struct 32:115–133. 5. Champoux JJ, Dulbecco R (1972) An activity from mammalian cells that untwists su- 28. Flynn RL, Zou L (2010) Oligonucleotide/oligosaccharide-binding fold proteins: A perhelical DNA: A possible swivel for DNA replication (polyoma-ethidium bromide- growing family of genome guardians. Crit Rev Biochem Mol Biol 45(4):266–275. 29. Anderson P, Kedersha N (2009) RNA granules: Post-transcriptional and epigenetic mouse-embryo cells-dye binding assay). Proc Natl Acad Sci USA 69(1):143–146. BIOCHEMISTRY – 6. Tse YC, Kirkegaard K, Wang JC (1980) Covalent bonds between protein and DNA. modulators of gene expression. Nat Rev Mol Cell Biol 10(6):430 436. Formation of phosphotyrosine linkage between certain DNA topoisomerases and 30. Salzman J, Gawad C, Wang PL, Lacayo N, Brown PO (2012) Circular RNAs are the DNA. J Biol Chem 255(12):5560–5565. predominant transcript isoform from hundreds of human genes in diverse cell types. 7. Schoeffler AJ, Berger JM (2008) DNA topoisomerases: Harnessing and constraining PLoS One 7(2):e30733. energy to govern chromosome topology. Q Rev Biophys 41(1):41–101. 31. Jeck WR, et al. (2013) Circular RNAs are abundant, conserved, and associated with – 8. Wang H, Di Gate RJ, Seeman NC (1996) An RNA topoisomerase. Proc Natl Acad Sci ALU repeats. RNA 19(2):141 157. USA 93(18):9477–9482. 32. Salzman J, Chen RE, Olsen MN, Wang PL, Brown PO (2013) Cell-type specific features 9. Sekiguchi J, Shuman S (1997) Site-specific ribonuclease activity of eukaryotic DNA of circular RNA expression. PLoS Genet 9(9):e1003777. 33. Rybak-Wolf A, et al. (2015) Circular RNAs in the mammalian brain are highly abun- topoisomerase I. Mol Cell 1(1):89–97. dant, conserved, and dynamically expressed. Mol Cell 58(5):870–885. 10. DiGate RJ, Marians KJ (1992) Escherichia coli topoisomerase III-catalyzed cleavage of 34. Westholm JO, et al. (2014) Genome-wide analysis of drosophila circular RNAs reveals RNA. J Biol Chem 267(29):20532–20535. their structural and sequence properties and age-dependent neural accumulation. 11. Stoll G, et al. (2013) Deletion of TOP3β, a component of FMRP-containing mRNPs, Cell Reports 9(5):1966–1980. contributes to neurodevelopmental disorders. Nat Neurosci 16(9):1228–1237. 35. You X, et al. (2015) Neural circular RNAs are derived from synaptic genes and regu- 12. Xu D, et al. (2013) Top3β is an RNA topoisomerase that works with fragile X syndrome lated by development and plasticity. Nat Neurosci 18(4):603–610. protein to promote synapse formation. Nat Neurosci 16(9):1238–1247. 36. Ashwal-Fluss R, et al. (2014) circRNA biogenesis competes with pre-mRNA splicing. 13. Yang Y, et al. (2014) Arginine methylation facilitates the recruitment of TOP3B to Mol Cell 56(1):55–66. chromatin to prevent R loop accumulation. Mol Cell 53(3):484–497. 37. Hansen TB, et al. (2013) Natural RNA circles function as efficient microRNA sponges. 14. Yang Y, et al. (2010) TDRD3 is an effector molecule for arginine-methylated histone Nature 495(7441):384–388. marks. Mol Cell 40(6):1016–1023. 38. Memczak S, et al. (2013) Circular RNAs are a large class of animal RNAs with regu- 15. Sims RJ, 3rd, et al. (2011) The C-terminal domain of RNA polymerase II is modified by latory potency. Nature 495(7441):333–338. – site-specific methylation. Science 332(6025):99 103. 39. Zhang Y, et al. (2013) Circular intronic long noncoding RNAs. Mol Cell 51(6):792–806. 16. Bhakar AL, Dölen G, Bear MF (2012) The pathophysiology of fragile X (and what it 40. Wilson-Sali T, Hsieh TS (2002) Generation of double-stranded breaks in hyper- – teaches us about synapses). Annu Rev Neurosci 35:417 443. negatively supercoiled DNA by Drosophila topoisomerase IIIbeta, a type IA enzyme. 17. Darnell JC, et al. (2011) FMRP stalls ribosomal translocation on mRNAs linked to J Biol Chem 277(30):26865–26871. – synaptic function and autism. Cell 146(2):247 261. 41. Wilson-Sali T, Hsieh TS (2002) Preferential cleavage of plasmid-based R-loops and 18. Wilson TM, Chen AD, Hsieh T (2000) Cloning and characterization of Drosophila D-loops by Drosophila topoisomerase IIIbeta. Proc Natl Acad Sci USA 99(12): topoisomerase IIIbeta. Relaxation of hypernegatively supercoiled DNA. J Biol Chem 7974–7979. – 275(3):1533 1540. 42. Thresher R, Griffith J (1992) Electron microscopic visualization of DNA and DNA- – 19. Cech TR, Bass BL (1986) Biological catalysis by RNA. Annu Rev Biochem 55:599 629. protein complexes as adjunct to biochemical studies. Methods Enzymol 211: – 20. Cech TR (1990) Self-splicing of group I introns. Annu Rev Biochem 59:543 568. 481–490. 21. Puttaraju M, Been MD (1992) Group I permuted intron-exon (PIE) sequences self-splice 43. Griffith JD, Christiansen G (1978) Electron microscope visualization of chromatin and to produce circular exons. Nucleic Acids Res 20(20):5357–5364. other DNA-protein complexes. Annu Rev Biophys Bioeng 7:19–35. 22. Arnott S, Hukins DW, Dover SD (1972) Optimised parameters for RNA double-helices. 44. Shaiu WL, Hsieh TS (1998) Targeting to transcriptionally active loci by the hydrophilic Biochem Biophys Res Commun 48(6):1392–1399. N-terminal domain of Drosophila DNA topoisomerase I. Mol Cell Biol 18(7):4358–4367. 23. Gast FU, Hagerman PJ (1991) Electrophoretic and hydrodynamic properties of duplex 45. Lee SH, Siaw GE, Willcox S, Griffith JD, Hsieh TS (2013) Synthesis and dissolution of ribonucleic acid molecules transcribed in vitro: Evidence that A-tracts do not generate hemicatenanes by type IA DNA topoisomerases. Proc Natl Acad Sci USA 110(38): curvature in RNA. Biochemistry 30(17):4268–4277. E3587–E3594. 24. Goulet I, Boisvenue S, Mokas S, Mazroui R, Côté J (2008) TDRD3, a novel Tudor do- 46. Puttaraju M, Beebe JA, Niranjanakumari S, Been MD, Fierke CA (1995) Generation main-containing protein, localizes to cytoplasmic stress granules. Hum Mol Genet and characterization of circular Bacillus subtilis RNase P RNA; activation by RNase P 17(19):3055–3074. protein. Nucleic Acids Symp Ser (33):92–94.

Siaw et al. PNAS | Published online August 31, 2016 | E5551 Downloaded by guest on October 2, 2021